Graphics Cards - Page 65

Sigh... there's another graphics card scam going around, this time with the flagship NVIDIA GeForce RTX 4090 which was found with a relabeled RTX 3080 Ti GPU inside. Check it out:

Northwestrepair revealed the scam in a new video, with what should've been a brand new ASUS TUF Gaming GeForce RTX 4090 graphics card, with the protective foil on it and all, but it was all a sc am. Once the TUF Gaming RTX 4090 was pulled apart, the GPU was labeled "AD102" but it was not the AD102 GPU, it was the GA102 GPU which powers the GeForce RTX 3080 Ti.

The two GPUs are nearly the same size: 628mm2 versus 608mm2, so it wasn't obvious at first, but it seems someone has polished the die to remove the label, and then laser-etched brand new ones... quite elaborate. Northwestrepair also noticed that at least one of the memory modules was not 2GB, but 256MB... and they weren't GDDR6X memory at all.

If you've seen one of Palit's over-the-top GameRock graphics cards, you can spot one from a mile away. Take the flagship Palit GeForce RTX 4090 GameRock OC as an example, where the physical design features a 'shimmering RGB crystal' look that is so gaudy you can't help but be impressed.

It's a design that has become synonymous with the GameRock sub-brand. However, Palit is about to change things up with a new GeForce RTX GameRock design slated to launch in the second half of 2024. Based on the timing, we wouldn't be surprised if this ends up being the GeForce RTX 5090 or GeForce RTX 5080 GameRock.

First, the new design is for fans of white-colored gear and those rocking white PC cases. Second, it removes the 'shimmering RGB crystal' look for a stylish wave of neon lights dancing around the three fans. I think it looks more appealing than the current crystal-heavy GameRock design, but hey, that's me.

ASRock has rebranded its Radeon 7900 WS series to the new Radeon 7900 Creator Series, which feature a blower-style cooler and newer 12V-2x6 power connector. Check them out:

The company previously had the Radeon RX 7900 XTX WS and Radeon RX 7900 XT WS, which have been rebranded to the Creator Series, designed for multi-GPU clusters for AI inference, generative AI, and professional workstation use.

ASRock's new cards feature a blower-style cooler, a nice-and-tight dual-slot design with a vapor chamber, and the newer 12V-2x6 power connector. I think the new 12V-2x6 power connector is the biggest news, as it seems ASRock is the first with a Radeon GPU using the new 16-pin PCIe power connector, which was exclusive to NVIDIA's current-gen GeForce RTX 40 series "Ampere" GPUs.

NVIDIA launched its SUPER refresh for the GeForce RTX 40 Series earlier this year, with the RTX 4070 SUPER, RTX 4070 Ti SUPER, and RTX 4080 SUPER. A GeForce RTX 4090 SUPER was absent, which would have introduced a new flagship beast.

With no sign of a GeForce RTX 4090 Ti, the current flagship will most likely remain as is until the arrival of the GeForce RTX 5090 later this year (or early 2025, depending on when NVIDIA decides to launch its next-gen Blackwell GPUs). However, thanks to the modding team at Teclab, we now glimpse what a GeForce RTX 4090 SUPER looks like - and it's impressive.

Teclab's creation is impressively 13% faster than the stock GeForce RTX 4090, even though it features the same GPU with 16384 CUDA Cores. It even managed a score almost 40% higher than the reference RTX 4090 in Unigine Superposition at 8K - however, that was with LOD tweaks inflating the result.

The Steam Hardware & Software Survey results for June 2024 are in, and the story remains reasonably close to what we saw in May 2024. NVIDIA sits pretty with a 76.38% share of GPUs used by PC gamers, followed by AMD with a 15.59% share. It's another bleak picture for Team Red, with Valve's data showing that there are more flagship GeForce RTX 4090 GPUs with PC gamers than any discrete Radeon GPU.

For the current generation of graphics cards, covering the GeForce RTX 40 Series, Radeon RX 7000 Series, and Intel Arc Graphics Series lineups, the only non-NVIDIA card to make the list is the Radeon RX 7900 XTX. As for Team Green, the Steam Hardware & Software Survey results for June 2024 show that the GeForce RTX 4060, RTX 4060 Ti, and RTX 4070 are very popular cards among PC gamers.

Here's a look at the Top 20 GPUs according to the latest Steam Hardware & Software Survey results for June 2024. Outside of a couple of integrated graphics options from AMD and Intel, the list includes cards from the current and previous three generations of GeForce hardware.

Micron announced nearly a month ago now that it was sampling its next-generation GDDR7 graphics memory, where we can expect 30% more performance in gaming at 1080p, 1440p, and 4K over GDDR6-based cards.

The US-based memory company teases that its new 32Gbps GDDR7 memory modules will provide up to 30% more performance in gaming -- in both ray tracing and rasterization -- compared to GDDR6 (to be specific, non-X) memory. Micron says its next-gen GDDR7 memory provides over 1.5TB/sec of memory bandwidth, which should have all gamers excited compared to the GDDR6X-powered GeForce RTX 4090 and its 1TB/sec memory bandwidth.

Micron's new GDDR7 memory with 32Gbps of bandwidth is 10Gbps high than the fastest GDDR6X memory modules on the market, with a test setup by Micron using 12 GDDR7 ICs connected to a 384-bit memory bus teasing a whopping 1.5TB/sec of memory bandwidth.

Jon Peddie Research provides informative quarterly data and industry analysis - covering shipments and quarterly trends. This year, we've shared some of the firm's data on 2024 GPU shipments and the overall market share for the big players in the space: NVIDIA, AMD, and Intel.

However, as an internet platform, Jon Peddie Research also posts news about the CPU and GPU industry and PC gaming technology. A recent post by the firm is worth checking out, mainly because of how funny and strange it is. Leading those who have spotted it to believe it's an attempt at AI-generated content.

And a failed one at that. The article, titled 'Leaked information reveals Intel GPU,' discusses a new graphics card from Team Blue that is in development - citing an 'unconfirmed source.' The nonsense tech the article mentions makes it sound like poorly generated AI content.

Intel's next-generation Arc Battlemage "X2" and "X3" graphics cards have been spotted, featuring up to 56 Xe2-GPU cores on a 256-bit memory bus.

We're hearing new information on Intel's next-gen Battlemage X2 and X3 GPUs from nbd.ltd and Volza, who have spotted the log in customs and shipping data on their respective online databases. Intel's new Arc Battlemage "X2" and "X3" could be codenames for next-gen Arc GPUs, or just internal codenames for the cards themselves.

There's another card that was listed as "Churchill Falls" and is reportedly a canceled SKU that's based on the Intel BMG-G10 GPU die. First off, we've got the Intel Arc Battlemage "X3," which uses the BMG-G10 GPU die, the flagship SKU from Intel that appears to be canceled, with the new BMG-G21 and BMG-G31 GPU dies taking its place.

ASUS has launched a new version of its Radeon RX 6500 XT DUAL OC V2 graphics card. The V2 has the same physical design and dimensions as its predecessor (201 x 128 x 40mm). However, there looks to be a different aluminum fin-stack heatsink layout and no RGB lighting this time around. As an RDNA 2-era GPU, it still sports the older DUAL design.

ASUS's Radeon RX 6500 XT DUAL OC V2 is a compact card two-slot card, so it wouldn't have any issues fitting in a small form-factor case or build. Elsewhere, the Radeon RX 6500 XT DUAL OC V2 graphics card features the same overclocked 2820 MHz boost clock as its predecessor, a notable increase over AMD's reference spec of 2610 MHz.

Unfortunately, the underlying memory configuration has not been upgraded, as we saw with ASRock's recent Radeon RX 6500 XT Phantom Gaming 8GB OC, launched a few weeks ago. As the naming suggests, it doubled the Radeon RX 6500 XT's VRAM capacity to 8GB compared to the 4GB of the reference design and ASUS's Radeon RX 6500 XT DUAL OC V2 graphics card.

At Siggraph 2023, NVIDIA announced its new AI-powered Neural Texture Compression (NTC) to reduce texture sizes without sacrificing detail. This would mean download and install sizes for games like Call of Duty or cinematic titles with optional 4K texture packs would decrease for PC gamers. This is excellent news in the age of games taking up 150 GB of storage space.

This week, we've learned that AMD is developing something similar, and it will be AI-powered. It's called Neural Texture Block Compression, and AMD will present its solution at the Eurographics Symposium on Rendering (EGSR) in London on July 2.

"Nobody likes downloading huge game packages," AMD's GPUOpen team writes in the announcement. "Our method compresses the texture using a neural network, reducing data size. Unchanged runtime execution allows easy game integration."

MSI introduced its new EXPERT series high-end graphics card design with the GeForce RTX 4080 SUPER EXPERT - with its minimal all-metal shroud and cooling design inspired by NVIDIA's Founders Edition line-up. It's a beautiful design we can't wait to see used again in future models; check out our full review for a breakdown of the design and performance.

Yesterday, images of a custom "Black Myth: Wukong" edition of the MSI GeForce RTX 4070 Ti SUPER EXPERT emerged (via Videocardz.com and @hongxing2020 on X), and it looked so good that many believed it to be the real deal. Unfortunately, you're looking at an impressive mod from X user @Lyrae622, who created the card for the highly anticipated game.

A custom Black Myth: Wukong graphics card makes sense, as the game features a suite of RTX technologies, including full ray-tracing DLSS Super Resolution and Frame Generation. Thanks to its impressive visuals, NVIDIA has promoted the game in the past, so we'd love to see MSI and NVIDIA reach out to @Lyrae622 to create a few of these to give away.

The NVIDIA GeForce RTX 4070 Ti SUPER was released earlier this year. It offers a minor performance bump over the previous non-SUPER variant and a notable increase in the VRAM capacity without a price increase. It launched with NVIDIA's AD103 GPU chip, also featured in the GeForce RTX 4080 and RTX 4080 SUPER.

The same GPU chips across various models are uncommon; it depends on the configuration. In the past week, we've seen several reports of new GeForce RTX 4070 Ti SUPER cards from companies like MSI, Colorful, and InnoVision using the 'AD102-225-A1' chip. AD102, for those unfamiliar, is the Ada Lovelace big boy found inside the GeForce RTX 4090.

It's found in the new MSI GeForce RTX 4070 Ti SUPER 16G Ventus 3X Black OC graphics card. However, this doesn't mean you'll get RTX 4090-like performance. It features the same CUDA Core count of 8448 and 16GB of GDDR6X memory on a 256-bit bus. With the full AD102 chip featuring 18,432 CUDA Cores, this new chip only has 45.8% of the SMs enabled.

GIGABYTE has just introduced its new Radeon PRO W7900 Dual Slot AI TOP 48G and Radeon PRO W7800 AI TOP 32G for workstation users, professionals, content creators, and AI developers. Check them out:

The new Radeon PRO W7900 Dual Slot AI TOP 48G has 48GB of GDDR6 memory, and is a smaller version of the original Radeon PRO W7900 workstation card. GIGABYTE is sticking pretty close to AMD's own reference design, with a black blower-style cooler and dual-slot design. There's a nice big "AI TOP" sticker on top, because this is an AI TOP graphics card.

Inside, the card has the Navi 31 GPU with 6144 shaders, and 48GB of GDDR6 memory with ECC support on a 384-bit memory bus with up to 864GB/sec of memory bandwidth.

GIGABYTE has just unleashed a new GeForce RTX 4070 Ti SUPER graphics card, with a hidden 16-pin PCIe power connector and a new WINDFORCE MAX cooling system.

The new GIGABYTE GeForce RTX 4070 Ti SUPER WINDFORCE MAX OC16 graphics card features factory overclocking out of the box, with the GPU boosting up to 2655MHz, which is a 45MHz increase over the reference RTX 4070 Ti SUPER.

Inside, we've got the regular 8448 CUDA cores from the AD103-275 GPU die, 16GB of GDDR6X memory on a 256-bit memory bus with up to 672.3GB/sec of memory bandwidth. We should have TGP ratings of close to the 285W reference form, too.

Back when EVGA made NVIDIA graphics cards, the firm produced certain high-end models with overclocker Kingpin - and these special edition cards are coming back to the GPU market, except with PNY this time.

Those who are familiar with the Kingpin brand will, of course, know that these are graphics cards built in conjunction with the high-profile overclocker of the same name (Vince 'Kingpin' Lucido).

As uncovered in an interview with Gamers Nexus (via VideoCardz), Kingpin's next move is to partner with PNY to produce graphics cards built for overclocking to fill the "huge hole in the enthusiast market."

The Minisforum DEG1 eGPU Dock is a very cool and minimal approach to hooking up an external GPU to a mini PC or laptop. It uses the OCuLink interface to deliver 64 Gbps of bandwidth, and the open-air design means you could hook up anything from a GeForce RTX 3060 to a GeForce RTX 4090 or the flagship Radeon RX 7900 XTX.

There's an included signal amplifier to ensure a stable connection, though the $99 USD price tag means you'll need a PC ATX or SFX power supply to provide enough power to the GPU. It's an elegant approach, but it does mean that you'd probably need to shell out for a power supply - and in that case, you'd want the most compact and modular option to help maintain a minimal look.

It's a niche product, but with the rise of portable PCs with powerful hardware-including PC gaming handhelds - there's a definite market for it. In fact, with a few mini PCs currently sitting in my living room, I wouldn't mind one of these to play more than Hades 2 on my big-screen OLED.

NVIDIA's RTX 4080 Super has been cut to its lowest price yet, with the GPU that was introduced at the start of 2024 now down to $949.

VideoCardz was on the ball enough to notice that at Newegg in the US, PNY's RTX 4080 Super Verto graphics card has had its price chopped again. Although you may notice the sticker price, as it were, is $979 - you need to apply a discount code at checkout which reduces it by a further $30 to get the board down to $949.

Previously, this PNY model has been below the MSRP of $999, and was spotted at $969 recently, but this is a fresh - and welcome - low for the cost of the Verto model.

Samsung Electronics is moving in a slightly different direction it seems, investing in GPU technology that would see it shift away from its typical memory semiconductor, and foundry services.

In a new report from Business Korea, Samsung Electronics' corporate governance report teases that the management committee ratified the "GPU Investment Proposal" back in March 2024. The committee includes senior executives including Han Jong-hee, the head of the Device eXperience (DX) division, the presidents of the Mobile Experience (MX) and Memory Business departments.

This is not the first meeting, but the third meeting of the year after the first GPU investment decision was publicly disclosed in 2012, which Business Korea reports is "sparking speculation that Samsung is poised to enhance its competitiveness in GPU-related business areas".

Regarding desktop GPUs for PC gaming, NVIDIA is dominating - a recent report by Jon Peddie Research (JPR) showed that 88% of AIB GPU shipments for Q1 2024 were GeForce RTX cards. Looking at the latest Steam Hardware & Software Survey results for May 2024, it's a similar picture: 76.18% of PC gamers across desktop rigs and laptops are playing with GeForce RTX cards.

Drilling down into the data and looking at the most popular GPUs, the picture becomes even worse for AMD and Intel. With 15.96% of the Steam GPU share, AMD's second-place position is bolstered by integrated Radeon graphics in mobile chips and gaming handhelds like Valve's Steam Deck. According to the latest data, the most popular discrete desktop gaming GPU is the Radeon RX 580, a GeForce GTX 1060 competitor from 2017.

With a 7.62% market share, all Intel entries in the list cover the company's integrated and aging Xe, UHD, and HD graphics technologies - there's no Arc mentioned anywhere. And when it comes to modern graphics cards, NVIDIA's latest GeForce RTX 40 Series is on its own.

Since it launched into Beta earlier his year, I switched to the new NVIDIA App for my main gaming rig and laptop, as the new interface and overlay are fast and feature-packed. At Computex 2024 the Control Panel and GeForce Experience replacement also got a meaty update in the form of 120 FPS ShadowPlay AV1 recording and one-click overclocking for GeForce RTX GPUs.

So yeah, there's no real reason to stick with the aging GeForce Experience app for GeForce RTX owners. However, seeing as the NVIDIA App is still opt-in and in Beta, Experience is still the official way to access GeForce RTX features like ShadowPlay game recording, performance monitoring tools, and optimized game settings. However, NVIDIA has almost forgotten that the GeForce Experience existed.

The new GeForce Experience 3.28 update is the app's first since November 2023. It brings a massive 122 'optimal game settings' for titles like Call of Duty: Modern Warfare III, Diablo IV, Alan Wake 2, Starfield, HELLDIVERS 2, Horizon Forbidden West, and many more.